changeset 95610:cb95dadcb2de

Merge from gnus--devo--0 Revision: emacs@sv.gnu.org/emacs--devo--0--patch-1215
author Miles Bader <miles@gnu.org>
date Fri, 06 Jun 2008 22:53:14 +0000
parents 31c3df6cae52
children 4ad2aacd19df
files doc/misc/ChangeLog doc/misc/gnus.texi lisp/ChangeLog lisp/Makefile.in lisp/gnus/ChangeLog lisp/gnus/nnir.el
diffstat 6 files changed, 2343 insertions(+), 5 deletions(-) [+]
line wrap: on
line diff
--- a/doc/misc/ChangeLog	Fri Jun 06 21:28:20 2008 +0000
+++ b/doc/misc/ChangeLog	Fri Jun 06 22:53:14 2008 +0000
@@ -1,3 +1,16 @@
+2008-06-05  Reiner Steib  <Reiner.Steib@gmx.de>
+
+	* gnus.texi (nnmairix): Markup and othe minor fixes.
+
+2008-06-05  David Engster  <dengste@eml.cc>
+
+	* gnus.texi (nnmairix): New nodes describing nnmairix.el.
+
+2008-06-05  Reiner Steib  <Reiner.Steib@gmx.de>
+
+	* gnus.texi (Group Parameters): Change ~/.gnus to ~/.gnus.el
+	(Searching, nnir, nnmairix): New stub nodes.
+
 2008-05-30  "Felix -"  <felix.1@canids.net>  (tiny change)
 
 	* cl.texi (Iteration Clauses): Fix incorrect "identical" examples.
@@ -42,10 +55,6 @@
 
 2008-04-13  Reiner Steib  <Reiner.Steib@gmx.de>
 
-	* gnus.texi (Searching, nnir.el, nnmairix.el): New nodes.
-
-2008-04-13  Reiner Steib  <Reiner.Steib@gmx.de>
-
 	* gnus.texi (Oort Gnus): Add message-fill-column.
 
 2008-04-12  Adrian Aichner  <adrian@xemacs.org>
--- a/doc/misc/gnus.texi	Fri Jun 06 21:28:20 2008 +0000
+++ b/doc/misc/gnus.texi	Fri Jun 06 22:53:14 2008 +0000
@@ -1736,6 +1736,7 @@
 * Exiting Gnus::                Stop reading news and get some work done.
 * Group Topics::                A folding group mode divided into topics.
 * Non-ASCII Group Names::       Accessing groups of non-English names.
+* Searching::                   Mail search engines.
 * Misc Group Stuff::            Other stuff that you can to do.
 @end menu
 
@@ -3167,7 +3168,7 @@
 question to @code{gnus-newsgroup-variables}.  @xref{Various Summary
 Stuff}.  So if you want to set @code{message-from-style} via the group
 parameters, then you may need the following statement elsewhere in your
-@file{~/.gnus} file:
+@file{~/.gnus.el} file:
 
 @lisp
 (add-to-list 'gnus-newsgroup-variables 'message-from-style)
@@ -4356,6 +4357,659 @@
 header will be displayed incorrectly in the article buffer.
 
 
+@node Searching
+@section Searching
+
+@menu
+* nnir::                     Searching on IMAP, with swish, namazu, etc.
+* nnmairix::                 Searching maildir, MH or mbox with Mairix.
+@end menu
+
+@cindex Searching
+
+FIXME: This node is a stub.
+
+FIXME: Add a brief overview of Gnus search capabilities.  A brief
+comparison of nnir, nnmairix, contrib/gnus-namazu would be nice
+as well.
+
+FIXME: Explain difference to @ref{Searching for Articles}, add reference
+and back-reference.
+
+@node nnir
+@subsection nnir
+
+FIXME: As a first step, convert the commentary of @file{nnir} to texi.
+@cindex nnir
+
+@node nnmairix
+@subsection nnmairix
+
+@cindex mairix 
+@cindex nnmairix
+This paragraph describes how to set up mairix and the back end
+@code{nnmairix} for indexing and searching your mail from within
+Gnus.  Additionally, you can create permanent ``smart'' groups which are
+bound to mairix searches and are automatically updated.
+
+@menu
+* About mairix::                About the mairix mail search engine
+* nnmairix requirements::       What you will need for using nnmairix
+* What nnmairix does::          What does nnmairix actually do?
+* Setting up mairix::           Set up your mairix installation
+* Configuring nnmairix::        Set up the nnmairix back end
+* nnmairix keyboard shortcuts:: List of available keyboard shortcuts
+* Propagating marks::           How to propagate marks from nnmairix groups
+* nnmairix tips and tricks::    Some tips, tricks and examples
+* nnmairix caveats::            Some more stuff you might want to know
+@end menu
+
+@c FIXME: The markup in this section needs improvement.  E.g. add
+@c @sample{...}, maybe remove some @strong{...}, convert ` - ' to `---`,
+@c ...
+
+@node About mairix
+@subsubsection About mairix
+
+Mairix is a tool for indexing and searching words in locally stored
+mail.  It was written by Richard Curnow and is licensed under the
+GPL.  Mairix comes with most popular GNU/Linux distributions, but it also
+runs under Windows (with cygwin), Mac OS X and Solaris.  The homepage can
+be found at
+@uref{http://www.rpcurnow.force9.co.uk/mairix/index.html}
+
+Though mairix might not be as flexible as other search tools like
+swish++ or namazu, which you can use via the @code{nnir} back end, it
+has the prime advantage of being incredibly fast.  On current systems, it
+can easily search through headers and message bodies of thousands and
+thousands of mails in well under a second.  Building the database
+necessary for searching might take a minute or two, but only has to be
+done once fully.  Afterwards, the updates are done incrementally and
+therefore are really fast, too.  Additionally, mairix is very easy to set
+up.
+
+For maximum speed though, mairix should be used with mails stored in
+@code{Maildir} or @code{MH} format (this includes the @code{nnml} back
+end), although it also works with mbox.  Mairix presents the search
+results by populating a @emph{virtual} maildir/MH folder with symlinks
+which point to the ``real'' message files (if mbox is used, copies are
+made).  Since mairix already presents search results in such a virtual
+mail folder, it is very well suited for using it as an external program
+for creating @emph{smart} mail folders, which represent certain mail
+searches.  This is similar to a Kiboze group (@pxref{Kibozed Groups}),
+but much faster.
+
+@node nnmairix requirements
+@subsubsection nnmairix requirements
+
+Mairix searches local mail - that means, mairix absolutely must have
+direct access to your mail folders.  If your mail resides on another
+server (e.g. an @acronym{IMAP} server) and you happen to have shell
+access, @code{nnmairix} supports running mairix remotely, e.g. via ssh.
+
+Additionally, @code{nnmairix} only supports the following Gnus back
+ends: @code{nnml}, @code{nnmaildir}, and @code{nnimap}.  You
+@strong{must} use one of these back ends for using
+@code{nnmairix}.  Other back ends, like @code{nnmbox}, @code{nnfolder} or
+@code{nnmh}, won't work.
+
+If you absolutely must use mbox and still want to use @code{nnmairix},
+you can set up a local @acronym{IMAP} server, which you then access via
+@code{nnimap}.  This is a rather massive setup for accessing some mbox
+files, so just change to MH or Maildir already...
+
+@node What nnmairix does
+@subsubsection What nnmairix does
+
+The back end @code{nnmairix} enables you to call mairix from within Gnus,
+either to query mairix with a search term or to update the
+database.  While visiting a message in the summary buffer, you can use
+several pre-defined shortcuts for calling mairix, e.g. to quickly
+search for all mails from the sender of the current message or to
+display the whole thread associated with the message, even if the
+mails are in different folders.
+
+Additionally, you can create permanent @code{nnmairix} groups which are bound
+to certain mairix searches.  This way, you can easily create a group
+containing mails from a certain sender, with a certain subject line or
+even for one specific thread based on the Message-ID.  If you check for
+new mail in these folders (e.g. by pressing @kbd{g} or @kbd{M-g}), they
+automatically update themselves by calling mairix. 
+
+You might ask why you need @code{nnmairix} at all, since mairix already
+creates the group, populates it with links to the mails so that you can
+then access it with Gnus, right? Well, this @emph{might} work, but often
+does not - at least not without problems.  Most probably you will get
+strange article counts, and sometimes you might see mails which Gnus
+claims have already been canceled and are inaccessible.  This is due to
+the fact that Gnus isn't really amused when things are happening behind
+its back.  Another problem can be the mail back end itself, e.g. if you
+use mairix with an @acronym{IMAP} server (I had Dovecot complaining
+about corrupt index files when mairix changed the contents of the search
+group).  Using @code{nnmairix} should circumvent these problems.
+
+@code{nnmairix} is not really a mail back end - it's actually more like a
+wrapper, sitting between a ``real'' mail back end where mairix stores the
+searches and the Gnus front end.  You can choose between three different
+mail back ends for the mairix folders: @code{nnml}, @code{nnmaildir} or
+@code{nnimap}.  @code{nnmairix} will call the mairix binary so that the
+search results are stored in folders named
+@code{zz_mairix-<NAME>-<NUMBER>} on this mail back end, but it will
+present these folders in the Gnus front end only with @code{<NAME>}.  You
+can use an existing mail back end where you already store your mail, but
+if you're uncomfortable with @code{nnmairix} creating new mail groups
+alongside your other mail, you can also create e.g. a new
+@code{nnmaildir} server exclusively for mairix.  However, a special case
+exists if you want to use mairix remotely on an IMAP server with
+@code{nnimap} - here the mairix folders and your other mail must be on
+the same @code{nnimap} back end.
+
+@node Setting up mairix
+@subsubsection Setting up mairix
+
+First: create a backup of your mail folders (@pxref{nnmairix caveats}).
+
+Setting up mairix is easy: simply create a @file{.mairixrc} file with
+(at least) the following entries:
+
+@example
+# Your Maildir/MH base folder
+base=~/Maildir
+@end example
+
+This is the base folder for your mails.  All the following paths are
+relative to this base folder.  If you want to use @code{nnmairix} with
+@code{nnimap}, this base path has to point to the mail path where the
+@acronym{IMAP} server stores the mail folders!
+
+@c FIXME: Add typical examples?
+@example
+maildir= ... your maildir folders which should be indexed ...
+mh= ... your nnml/mh folders which should be indexed ...
+mbox = ... your mbox files which should be indexed ...
+@end example
+
+Specify all your maildir/nnml folders and mbox files (relative to the
+base path!) you want to index with mairix.  See the man-page for
+mairixrc for details.
+
+@example
+omit=zz_mairix-*
+@end example
+
+@vindex nnmairix-group-prefix
+This should make sure that you don't accidentally index the mairix
+search results.  You can change the prefix of these folders with the
+variable @code{nnmairix-group-prefix}.
+
+@c FIXME: Add typical examples?
+@example
+mformat= ... 'maildir' or 'mh' ...
+database= ... location of database file ...
+@end example
+
+The @code{format} setting specifies the output format for the mairix
+search folder.  Set this to @code{mh} if you want to access search results
+with @code{nnml}.  Otherwise choose @code{maildir}.
+
+See the man pages for mairix and mairixrc for further options.  Now
+simply call @code{mairix} to create the index for the first time.
+
+@node Configuring nnmairix
+@subsubsection Configuring nnmairix
+
+In group mode, type @kbd{G b c}
+(@code{nnmairix-create-server-and-default-group}).  This will ask you for all
+necessary information and create a @code{nnmairix} server as a foreign
+server.  You will have to specify the following:
+
+@itemize @bullet
+
+@item
+The @strong{name} of the @code{nnmairix} server - choose whatever you
+want.
+
+@item
+The @strong{mail back end} where mairix should stores its
+searches.  Currently @code{nnmaildir}, @code{nnimap} and @code{nnml} are
+supported.  As explained above, for locally stored mails, this can be an
+existing mail back end where you store your mails.  However, you can also
+create e.g. a new @code{nnmaildir} server exclusively for
+@code{nnmairix} in your secondary select methods (@pxref{Finding the
+News}).  If you want to use mairix remotely on an @acronym{IMAP} server,
+you have to choose the corresponding @code{nnimap} back end here.
+
+@item
+@vindex nnmairix-mairix-search-options
+The @strong{command} to call the mairix binary.  This will usually just
+be @code{mairix}, but you can also choose something like @code{ssh
+SERVER mairix} if you want to call mairix remotely, e.g. on your
+@acronym{IMAP} server.  If you want to add some default options to
+mairix, you could do this here, but better use the variable
+@code{nnmairix-mairix-search-options} instead.
+
+@item
+The name of the @strong{default search group}.  This will be the group
+where all temporary mairix searches are stored, i.e. all searches which
+are not bound to permanent @code{nnmairix} groups.  Choose whatever you
+like.
+
+@item
+If the mail back end is @code{nnimap} or @code{nnmaildir}, you will be
+asked if you work with @strong{Maildir++}, i.e. with hidden maildir
+folders (=beginning with a dot).  For example, you have to answer
+@samp{yes} here if you work with the Dovecot @acronym{IMAP}
+server.  Otherwise, you should answer @samp{no} here.
+
+@end itemize
+
+@node nnmairix keyboard shortcuts
+@subsubsection nnmairix keyboard shortcuts
+
+In group mode:
+
+@table @kbd
+
+@item G b c
+@kindex G b c (Group)
+@findex nnmairix-create-server-and-default-group
+Creates @code{nnmairix} server and default search group for this server
+(@code{nnmairix-create-server-and-default-group}).  You should have done
+this by now (@pxref{Configuring nnmairix}).
+
+@item G b s
+@kindex G b s (Group)
+@findex nnmairix-search
+Prompts for query which is then sent to the mairix binary.  Search
+results are put into the default search group which is automatically
+displayed (@code{nnmairix-search}).
+
+@item G b m
+@kindex G b m (Group)
+@findex nnmairix-widget-search
+Allows you to create a mairix search or a permanent group more
+comfortably using graphical widgets, similar to a customization
+group.  Just try it to see how it works (@code{nnmairix-widget-search}).
+
+@item G b i
+@kindex G b i (Group)
+@findex nnmairix-search-interactive
+Another command for creating a mairix query more comfortably, but uses
+only the minibuffer (@code{nnmairix-search-interactive}).
+
+@item G b g
+@kindex G b g (Group)
+@findex nnmairix-create-search-group
+Creates a permanent group which is associated with a search query
+(@code{nnmairix-create-search-group}).  The @code{nnmairix} back end
+automatically calls mairix when you update this group with @kbd{g} or
+@kbd{M-g}.
+
+@item G b q
+@kindex G b q (Group)
+@findex nnmairix-group-change-query-this-group
+Changes the search query for the @code{nnmairix} group under cursor
+(@code{nnmairix-group-change-query-this-group}).
+
+@item G b t
+@kindex G b t (Group)
+@findex nnmairix-group-toggle-threads-this-group
+Toggles the 'threads' parameter for the @code{nnmairix} group under cursor,
+i.e.  if you want see the whole threads of the found messages
+(@code{nnmairix-group-toggle-threads-this-group}).
+
+@item G b u
+@kindex G b u (Group)
+@findex nnmairix-update-database
+@vindex nnmairix-mairix-update-options
+Calls mairix binary for updating the database
+(@code{nnmairix-update-database}).  The default parameters are @code{-F}
+and @code{-Q} for making this as fast as possible (see variable
+@code{nnmairix-mairix-update-options} for defining these default
+options).
+
+@item G b r
+@kindex G b r (Group)
+@findex nnmairix-group-toggle-readmarks-this-group
+Keep articles in this @code{nnmairix} group always read or unread, or leave the
+marks unchanged (@code{nnmairix-group-toggle-readmarks-this-group}).
+
+@item G b d
+@kindex G b d (Group)
+@findex nnmairix-group-delete-recreate-this-group
+Recreate @code{nnmairix} group on the ``real'' mail back end
+(@code{nnmairix-group-delete-recreate-this-group}).  You can do this if
+you always get wrong article counts with a @code{nnmairix} group.
+
+@item G b a
+@kindex G b a (Group)
+@findex nnmairix-group-toggle-allowfast-this-group
+Toggles the @code{allow-fast} parameters for group under cursor
+(@code{nnmairix-group-toggle-allowfast-this-group}).  The default
+behavior of @code{nnmairix} is to do a mairix search every time you
+update or enter the group.  With the @code{allow-fast} parameter set,
+mairix will only be called when you explicitly update the group, but not
+upon entering.  This makes entering the group faster, but it may also
+lead to dangling symlinks if something changed between updating and
+entering the group which is not yet in the mairix database.
+
+@item G b p
+@kindex G b p (Group)
+@findex nnmairix-group-toggle-propmarks-this-group
+Toggle marks propagation for this group
+(@code{nnmairix-group-toggle-propmarks-this-group}).  (@pxref{Propagating
+marks}).
+
+@item G b o
+@kindex G b o (Group)
+@findex nnmairix-propagate-marks
+Manually propagate marks (@code{nnmairix-propagate-marks}); needed only when
+@code{nnmairix-propagate-marks-upon-close} is set to @code{nil}.
+
+@end table
+
+In summary mode:
+
+@table @kbd
+
+@item $ m
+@kindex $ m (Summary)
+@findex nnmairix-widget-search-from-this-article
+Allows you to create a mairix query or group based on the current
+message using graphical widgets (same as @code{nnmairix-widget-search})
+(@code{nnmairix-widget-search-from-this-article}).
+
+@item $ g
+@kindex $ g (Summary)
+@findex nnmairix-create-search-group-from-message
+Interactively creates a new search group with query based on the current
+message, but uses the minibuffer instead of graphical widgets
+(@code{nnmairix-create-search-group-from-message}).
+
+@item $ t
+@kindex $ t (Summary)
+@findex nnmairix-search-thread-this-article
+Searches thread for the current article
+(@code{nnmairix-search-thread-this-article}).  This is effectively a
+shortcut for calling @code{nnmairix-search} with @samp{m:msgid} of the
+current article and enabled threads.
+
+@item $ f
+@kindex $ f (Summary)
+@findex nnmairix-search-from-this-article
+Searches all messages from sender of the current article
+(@code{nnmairix-search-from-this-article}).  This is a shortcut for
+calling @code{nnmairix-search} with @samp{f:From}.
+
+@item $ o
+@kindex $ o (Summary)
+@findex nnmairix-goto-original-article
+(Only in @code{nnmairix} groups!) Tries determine the group this article
+originally came from and displays the article in this group, so that
+e.g. replying to this article the correct posting styles/group
+parameters are applied (@code{nnmairix-goto-original-article}).  This
+function will use the registry if available, but can also parse the
+article file path as a fallback method.
+
+@item $ u
+@kindex $ u (Summary)
+@findex nnmairix-remove-tick-mark-original-article
+Remove possibly existing tick mark from original article
+(@code{nnmairix-remove-tick-mark-original-article}).  (@pxref{nnmairix
+tips and tricks}).
+
+@end table
+
+@node Propagating marks
+@subsubsection Propagating marks
+
+First of: you really need a patched mairix binary for using the marks
+propagation feature efficiently. Otherwise, you would have to update
+the mairix database all the time. You can get the patch at
+
+@uref{http://m61s02.vlinux.de/mairix-maildir-patch.tar}
+
+You need the mairix v0.21 source code for this patch; everything else
+is explained in the accompanied readme file. If you don't want to use
+marks propagation, you don't have to apply these patches, but they also
+fix some annoyances regarding changing maildir flags, so it might still
+be useful to you.
+
+With the patched mairix binary, you can use @code{nnmairix} as an
+alternative to mail splitting (@pxref{Fancy Mail Splitting}). For
+example, instead of splitting all mails from @samp{david@@foobar.com}
+into a group, you can simply create a search group with the query
+@samp{f:david@@foobar.com}. This is actually what ``smart folders'' are
+all about: simply put everything in one mail folder and dynamically
+create searches instead of splitting. This is more flexible, since you
+can dynamically change your folders any time you want to. This also
+implies that you will usually read your mails in the @code{nnmairix}
+groups instead of your ``real'' mail groups.
+
+There is one problem, though: say you got a new mail from
+@samp{david@@foobar.com} - it will now show up in two groups, the
+``real'' group (your INBOX, for example) and in the @code{nnmairix}
+search group (provided you have updated the mairix database). Now you
+enter the @code{nnmairix} group and read the mail. The mail will be
+marked as read, but only in the @code{nnmairix} group - in the ``real''
+mail group it will be still shown as unread.
+
+You could now catch up the mail group (@pxref{Group Data}), but this is
+tedious and error prone, since you may overlook mails you don't have
+created @code{nnmairix} groups for. Of course, you could first use
+@code{nnmairix-goto-original-article} (@pxref{nnmairix keyboard
+shortcuts}) and then read the mail in the original group, but that's
+even more cumbersome.
+
+Clearly, the easiest way would be if marks could somehow be
+automatically set for the original article. This is exactly what
+@emph{marks propagation} is about.
+
+Marks propagation is deactivated by default. You can activate it for a
+certain @code{nnmairix} group with
+@code{nnmairix-group-toggle-propmarks-this-group} (bound to @kbd{G b
+p}). This function will warn you if you try to use it with your default
+search group; the reason is that the default search group is used for
+temporary searches, and it's easy to accidentally propagate marks from
+this group. However, you can ignore this warning if you really want to.
+
+With marks propagation enabled, all the marks you set in a @code{nnmairix}
+group should now be propagated to the original article. For example,
+you can now tick an article (by default with @kbd{!}) and this mark should
+magically be set for the original article, too.
+
+A few more remarks which you may or may not want to know:
+
+@vindex nnmairix-propagate-marks-upon-close
+Marks will not be set immediately, but only upon closing a group. This
+not only makes marks propagation faster, it also avoids problems with
+dangling symlinks when dealing with maildir files (since changing flags
+will change the file name). You can also control when to propagate marks
+via @code{nnmairix-propagate-marks-upon-close} (see the doc-string for
+details).
+
+Obviously, @code{nnmairix} will have to look up the original group for every
+article you want to set marks for. If available, @code{nnmairix} will first use
+the registry for determining the original group. The registry is very
+fast, hence you should really, really enable the registry when using
+marks propagation. If you don't have to worry about RAM and disc space,
+set @code{gnus-registry-max-entries} to a large enough value; to be on
+the safe side, choose roughly the amount of mails you index with mairix.
+
+@vindex nnmairix-only-use-registry
+If you don't want to use the registry or the registry hasn't seen the
+original article yet, @code{nnmairix} will use an additional mairix search
+for determining the file path of the article. This, of course, is way
+slower than the registry - if you set hundreds or even thousands of
+marks this way, it might take some time. You can avoid this situation
+by setting @code{nnmairix-only-use-registry} to t.
+
+Maybe you also want to propagate marks the other way round, i.e. if you
+tick an article in a "real" mail group, you'd like to have the same
+article in a @code{nnmairix} group ticked, too. For several good
+reasons, this can only be done efficiently if you use maildir. To
+immediately contradict myself, let me mention that it WON'T work with
+@code{nnmaildir}, since @code{nnmaildir} stores the marks externally and
+not in the file name. Therefore, propagating marks to @code{nnmairix}
+groups will usually only work if you use an IMAP server which uses
+maildir as its file format.
+
+@vindex nnmairix-propagate-marks-to-nnmairix-groups
+If you work with this setup, just set
+@code{nnmairix-propagate-marks-to-nnmairix-groups} to @code{t} and see what
+happens. If you don't like what you see, just set it to @code{nil} again. One
+problem might be that you get a wrong number of unread articles; this
+usually happens when you delete or expire articles in the original
+groups. When this happens, you can recreate the @code{nnmairix} group on the
+back end using @kbd{G b d}.
+
+@node nnmairix tips and tricks
+@subsubsection nnmairix tips and tricks
+
+@itemize
+@item
+Checking Mail
+
+@findex nnmairix-update-groups
+I put all my important mail groups at group level 1. The mairix groups
+have group level 5, so they do not get checked at start up (@pxref{Group
+Levels}).
+
+I use the following to check for mails:
+
+@lisp
+(defun my-check-mail-mairix-update (level)
+  (interactive "P")
+  ;; if no prefix given, set level=1
+  (gnus-group-get-new-news (or level 1))
+  (nnmairix-update-groups "mairixsearch" t t)
+  (gnus-group-list-groups))
+
+(define-key gnus-group-mode-map "g" 'my-check-mail-mairix-update)
+@end lisp
+
+Instead of @samp{"mairixsearch"} use the name of your @code{nnmairix}
+server. See the doc string for @code{nnmairix-update-groups} for
+details.
+
+@item
+Example: search group for ticked articles
+
+For example, you can create a group for all ticked articles, where the
+articles always stay unread:
+
+Hit @kbd{G b g}, enter group name (e.g. @samp{important}), use
+@samp{F:f} as query and do not include threads.
+
+Now activate marks propagation for this group by using @kbd{G b p}. Then
+activate the always-unread feature by using @kbd{G b r} twice.
+
+So far so good - but how do you remove the tick marks in the @code{nnmairix}
+group?  There are two options: You may simply use
+@code{nnmairix-remove-tick-mark-original-article} (bound to @kbd{$ u}) to remove
+tick marks from the original article. The other possibility is to set
+@code{nnmairix-propagate-marks-to-nnmairix-groups} to @code{t}, but see the above
+comments about this option.  If it works for you, the tick marks should
+also exist in the @code{nnmairix} group and you can remove them as usual,
+e.g. by marking an article as read.
+
+When you have removed a tick mark from the original article, this
+article should vanish from the @code{nnmairix} group after you have updated the
+mairix database and updated the group.  Fortunately, there is a function
+for doing exactly that: @code{nnmairix-update-groups}. See the previous code
+snippet and the doc string for details.
+
+@item
+Dealing with auto-subscription of mail groups
+
+As described before, all @code{nnmairix} groups are in fact stored on
+the mail back end in the form @samp{zz_mairix-<NAME>-<NUMBER>}. You can
+see them when you enter the back end server in the server buffer. You
+should not subscribe these groups! Unfortunately, these groups will
+usually get @strong{auto-subscribed} when you use @code{nnmaildir} or
+@code{nnml}, i.e. you will suddenly see groups of the form
+@samp{zz_mairix*} pop up in your group buffer. If this happens to you,
+simply kill these groups with C-k.  For avoiding this, turn off
+auto-subscription completely by setting the variable
+@code{gnus-auto-subscribed-groups} to @code{nil} (@pxref{Filtering New
+Groups}), or if you like to keep this feature use the following kludge
+for turning it off for all groups beginning with @samp{zz_}:
+
+@lisp
+(setq gnus-auto-subscribed-groups
+      "^\\(nnml\\|nnfolder\\|nnmbox\\|nnmh\\|nnbabyl\\|nnmaildir\\).*:\\([^z]\\|z$\\|\\z[^z]\\|zz$\\|zz[^_]\\|zz_$\\).*")
+@end lisp
+
+@end itemize
+
+@node nnmairix caveats
+@subsubsection nnmairix caveats
+
+@itemize
+@item
+If you use the Gnus agent (@pxref{Gnus Unplugged}): don't agentize
+@code{nnmairix} groups (though I have no idea what happens if you do).
+
+@item
+If you use the Gnus registry: don't use the registry with
+@code{nnmairix} groups (put them in
+@code{gnus-registry-unfollowed-groups}).  Be @strong{extra careful} if
+you use @code{gnus-registry-split-fancy-with-parent} - mails which are
+split into @code{nnmairix} groups are usually gone for good as soon as
+you check the group for new mail (yes, it has happened to me...).
+
+@item
+Therefore: @strong{Never ever} put ``real'' mails into @code{nnmairix}
+groups (you shouldn't be able to, anyway).
+
+@item
+mairix does only support us-ascii characters.
+
+@item
+@code{nnmairix} uses a rather brute force method to force Gnus to
+completely reread the group on the mail back end after mairix was
+called - it simply deletes and re-creates the group on the mail
+back end. So far, this has worked for me without any problems, and I
+don't see how @code{nnmairix} could delete other mail groups than its
+own, but anyway: you really should have a backup of your mail
+folders.
+
+@item
+All necessary information is stored in the group parameters
+(@pxref{Group Parameters}). This has the advantage that no active file
+is needed, but also implies that when you kill a @code{nnmairix} group,
+it is gone for good.
+
+@item
+@findex nnmairix-purge-old-groups
+If you create and kill a lot of @code{nnmairix} groups, the
+``zz_mairix-*'' groups will accumulate on the mail back end server. To
+delete old groups which are no longer needed, call
+@code{nnmairix-purge-old-groups}. Note that this assumes that you don't
+save any ``real'' mail in folders of the form
+@code{zz_mairix-<NAME>-<NUMBER>}. You can change the prefix of
+@code{nnmairix} groups by changing the variable
+@code{nnmairix-group-prefix}.
+
+@item
+The following only applies if you @strong{don't} use the mentioned patch
+for mairix (@pxref{Propagating marks}):
+
+A problem can occur when using @code{nnmairix} with maildir folders and
+comes with the fact that maildir stores mail flags like @samp{Seen} or
+@samp{Replied} by appending chars @samp{S} and @samp{R} to the message
+file name, respectively. This implies that currently you would have to
+update the mairix database not only when new mail arrives, but also when
+mail flags are changing. The same applies to new mails which are indexed
+while they are still in the @samp{new} folder but then get moved to
+@samp{cur} when Gnus has seen the mail. If you don't update the database
+after this has happened, a mairix query can lead to symlinks pointing to
+non-existing files. In Gnus, these messages will usually appear with
+``(none)'' entries in the header and can't be accessed. If this happens
+to you, using @kbd{G b u} and updating the group will usually fix this.
+
+@end itemize
+
 @node Misc Group Stuff
 @section Misc Group Stuff
 
--- a/lisp/ChangeLog	Fri Jun 06 21:28:20 2008 +0000
+++ b/lisp/ChangeLog	Fri Jun 06 22:53:14 2008 +0000
@@ -1,3 +1,7 @@
+2008-06-06  Miles Bader  <miles@gnu.org>
+
+	* Makefile.in (ELCFILES): Add gnus/nndir.elc.
+
 2008-06-06  Chong Yidong  <cyd@stupidchicken.com>
 
 	* menu-bar.el (menu-bar-options-menu): Add Menu entry for
--- a/lisp/Makefile.in	Fri Jun 06 21:28:20 2008 +0000
+++ b/lisp/Makefile.in	Fri Jun 06 22:53:14 2008 +0000
@@ -600,6 +600,7 @@
 	$(lisp)/gnus/nngateway.elc \
 	$(lisp)/gnus/nnheader.elc \
 	$(lisp)/gnus/nnimap.elc \
+	$(lisp)/gnus/nnir.elc \
 	$(lisp)/gnus/nnkiboze.elc \
 	$(lisp)/gnus/nnlistserv.elc \
 	$(lisp)/gnus/nnmail.elc \
--- a/lisp/gnus/ChangeLog	Fri Jun 06 21:28:20 2008 +0000
+++ b/lisp/gnus/ChangeLog	Fri Jun 06 22:53:14 2008 +0000
@@ -1,3 +1,7 @@
+2008-06-05  Reiner Steib  <Reiner.Steib@gmx.de>
+
+	* nnir.el: New file.
+
 2008-06-05  Stefan Monnier  <monnier@iro.umontreal.ca>
 
 	* gnus-util.el (gnus-read-shell-command): New function.
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/lisp/gnus/nnir.el	Fri Jun 06 22:53:14 2008 +0000
@@ -0,0 +1,1666 @@
+;;; nnir.el --- search mail with various search engines -*- coding: iso-8859-1 -*-
+
+;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004,
+;;   2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+
+;; Author: Kai Großjohann <grossjohann@ls6.cs.uni-dortmund.de>
+;; Swish-e and Swish++ backends by:
+;;   Christoph Conrad <christoph.conrad@gmx.de>.
+;; IMAP backend by: Simon Josefsson <jas@pdc.kth.se>.
+;; IMAP search by: Torsten Hilbrich <torsten.hilbrich <at> gmx.net>
+;; IMAP search improved by Daniel Pittman  <daniel@rimspace.net>.
+;; nnmaildir support for Swish++ and Namazu backends by:
+;;   Justus Piater <Justus <at> Piater.name>
+
+;; TODO: Documentation in the Gnus manual
+
+;; From: Reiner Steib
+;; Subject: Re: Including nnir.el
+;; Newsgroups: gmane.emacs.gnus.general
+;; Message-ID: <v9d5dnp6aq.fsf@marauder.physik.uni-ulm.de>
+;; Date: 2006-06-05 22:49:01 GMT
+;;
+;; On Sun, Jun 04 2006, Sascha Wilde wrote:
+;;
+;; > The one thing most hackers like to forget: Documentation.  By now the
+;; > documentation is only in the comments at the head of the source, I
+;; > would use it as basis to cook up some minimal texinfo docs.
+;; >
+;; > Where in the existing gnus manual would this fit best?
+
+;; Maybe (info "(gnus)Combined Groups") for a general description.
+;; `gnus-group-make-nnir-group' might be described in (info
+;; "(gnus)Foreign Groups") as well.
+
+;; Keywords: news mail searching ir
+
+;; This file is part of GNU Emacs.
+
+;; This is free software; you can redistribute it and/or modify
+;; it under the terms of the GNU General Public License as published by
+;; the Free Software Foundation; either version 3, or (at your option)
+;; any later version.
+
+;; GNU Emacs is distributed in the hope that it will be useful,
+;; but WITHOUT ANY WARRANTY; without even the implied warranty of
+;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+;; GNU General Public License for more details.
+
+;; You should have received a copy of the GNU General Public License
+;; along with GNU Emacs; see the file COPYING.  If not, write to the
+;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
+;; Boston, MA 02110-1301, USA.
+
+;;; Commentary:
+
+;; The most recent version of this can always be fetched from the Gnus
+;; CVS repository.  See http://www.gnus.org/ for more information.
+
+;; This code is still in the development stage but I'd like other
+;; people to have a look at it.  Please do not hesitate to contact me
+;; with your ideas.
+
+;; What does it do?  Well, it allows you to index your mail using some
+;; search engine (freeWAIS-sf, swish-e and others -- see later),
+;; then type `G G' in the Group buffer and issue a query to the search
+;; engine.  You will then get a buffer which shows all articles
+;; matching the query, sorted by Retrieval Status Value (score).
+
+;; When looking at the retrieval result (in the Summary buffer) you
+;; can type `G T' (aka M-x gnus-summary-nnir-goto-thread RET) on an
+;; article.  You will be teleported into the group this article came
+;; from, showing the thread this article is part of.  (See below for
+;; restrictions.)
+
+;; The Lisp installation is simple: just put this file on your
+;; load-path, byte-compile it, and load it from ~/.gnus or something.
+;; This will install a new command `G G' in your Group buffer for
+;; searching your mail.  Note that you also need to configure a number
+;; of variables, as described below.
+
+;; Restrictions:
+;;
+;; * If you don't use HyREX as your search engine, this expects that
+;;   you use nnml or another one-file-per-message backend, because the
+;;   others doesn't support nnfolder.
+;; * It can only search the mail backend's which are supported by one
+;;   search engine, because of different query languages.
+;; * There are restrictions to the Wais setup.
+;; * There are restrictions to the imap setup.
+;; * gnus-summary-nnir-goto-thread: Fetches whole group first, before
+;;   limiting to the right articles.  This is much too slow, of
+;;   course.  May issue a query for number of articles to fetch; you
+;;   must accept the default of all articles at this point or things
+;;   may break.
+
+;; The Lisp setup involves setting a few variables and setting up the
+;; search engine. You can define the variables in the server definition
+;; like this :
+;;   (setq gnus-secondary-select-methods '(
+;;       (nnimap "" (nnimap-address "localhost")
+;;                  (nnir-search-engine hyrex)
+;;                  (nnir-hyrex-additional-switches ("-d" "ddl-nnimap.xml"))
+;;       )))
+;; Or you can define the global ones. The variables set in the mailer-
+;; definition will be used first.
+;; The variable to set is `nnir-search-engine'.  Choose one of the engines
+;; listed in `nnir-engines'.  (Actually `nnir-engines' is an alist,
+;; type `C-h v nnir-engines RET' for more information; this includes
+;; examples for setting `nnir-search-engine', too.)
+;;
+;; The variable nnir-mail-backend isn't used anymore.
+;;
+
+;; You must also set up a search engine.  I'll tell you about the two
+;; search engines currently supported:
+
+;; 1. freeWAIS-sf
+;;
+;; As always with freeWAIS-sf, you need a so-called `format file'.  I
+;; use the following file:
+;;
+;; ,-----
+;; | # Kai's format file for freeWAIS-sf for indexing mails.
+;; | # Each mail is in a file, much like the MH format.
+;; |
+;; | # Document separator should never match -- each file is a document.
+;; | record-sep: /^@this regex should never match@$/
+;; |
+;; | # Searchable fields specification.
+;; |
+;; | region: /^[sS]ubject:/ /^[sS]ubject: */
+;; |         subject "Subject header" stemming TEXT BOTH
+;; | end: /^[^ \t]/
+;; |
+;; | region: /^([tT][oO]|[cC][cC]):/ /^([tT][oO]|[cC][cC]): */
+;; |         to "To and Cc headers" SOUNDEX BOTH
+;; | end: /^[^ \t]/
+;; |
+;; | region: /^[fF][rR][oO][mM]:/ /^[fF][rR][oO][mM]: */
+;; |         from "From header" SOUNDEX BOTH
+;; | end: /^[^ \t]/
+;; |
+;; | region: /^$/
+;; |         stemming TEXT GLOBAL
+;; | end: /^@this regex should never match@$/
+;; `-----
+;;
+;; 1998-07-22: waisindex would dump core on me for large articles with
+;; the above settings.  I used /^$/ as the end regex for the global
+;; field.  That seemed to work okay.
+
+;; There is a Perl module called `WAIS.pm' which is available from
+;; CPAN as well as ls6-ftp.cs.uni-dortmund.de:/pub/wais/Perl.  This
+;; module comes with a nifty tool called `makedb', which I use for
+;; indexing.  Here's my `makedb.conf':
+;;
+;; ,-----
+;; | # Config file for makedb
+;; |
+;; | # Global options
+;; | waisindex = /usr/local/bin/waisindex
+;; | wais_opt  = -stem -t fields
+;; | # `-stem' option necessary when `stemming' is specified for the
+;; | # global field in the *.fmt file
+;; |
+;; | # Own variables
+;; | homedir = /home/kai
+;; |
+;; | # The mail database.
+;; | database        = mail
+;; | files           = `find $homedir/Mail -name \*[0-9] -print`
+;; | dbdir           = $homedir/.wais
+;; | limit           = 100
+;; `-----
+;;
+;; The Lisp setup involves the `nnir-wais-*' variables.  The most
+;; difficult to understand variable is probably
+;; `nnir-wais-remove-prefix'.  Here's what it does: the output of
+;; `waissearch' basically contains the file name and the (full)
+;; directory name.  As Gnus works with group names rather than
+;; directory names, the directory name is transformed into a group
+;; name as follows: first, a prefix is removed from the (full)
+;; directory name, then all `/' are replaced with `.'.  The variable
+;; `nnir-wais-remove-prefix' should contain a regex matching exactly
+;; this prefix.  It defaults to `$HOME/Mail/' (note the trailing
+;; slash).
+
+;; 2. Namazu
+;;
+;; The Namazu backend requires you to have one directory containing all
+;; index files, this is controlled by the `nnir-namazu-index-directory'
+;; variable.  To function the `nnir-namazu-remove-prefix' variable must
+;; also be correct, see the documentation for `nnir-wais-remove-prefix'
+;; above.
+;;
+;; It is particularly important not to pass any any switches to namazu
+;; that will change the output format.  Good switches to use include
+;; `--sort', `--ascending', `--early' and `--late'.  Refer to the Namazu
+;; documentation for further information on valid switches.
+;;
+;; To index my mail with the `mknmz' program I use the following
+;; configuration file:
+;;
+;; ,----
+;; | package conf;  # Don't remove this line!
+;; |
+;; | # Paths which will not be indexed. Don't use `^' or `$' anchors.
+;; | $EXCLUDE_PATH = "spam|sent";
+;; |
+;; | # Header fields which should be searchable. case-insensitive
+;; | $REMAIN_HEADER = "from|date|message-id|subject";
+;; |
+;; | # Searchable fields. case-insensitive
+;; | $SEARCH_FIELD = "from|date|message-id|subject";
+;; |
+;; | # The max length of a word.
+;; | $WORD_LENG_MAX = 128;
+;; |
+;; | # The max length of a field.
+;; | $MAX_FIELD_LENGTH = 256;
+;; `----
+;;
+;; My mail is stored in the directories ~/Mail/mail/, ~/Mail/lists/ and
+;; ~/Mail/archive/, so to index them I go to the directory set in
+;; `nnir-namazu-index-directory' and issue the following command.
+;;
+;;      mknmz --mailnews ~/Mail/archive/ ~/Mail/mail/ ~/Mail/lists/
+;;
+;; For maximum searching efficiency I have a cron job set to run this
+;; command every four hours.
+
+;; 3. HyREX
+;;
+;; The HyREX backend requires you to have one directory from where all
+;; your relative paths are to, if you use them. This directory must be
+;; set in the `nnir-hyrex-index-directory' variable, which defaults to
+;; your home directory. You must also pass the base, class and
+;; directory options or simply your dll to the `nnir-hyrex-programm' by
+;; setting the `nnir-hyrex-additional-switches' variable accordently.
+;; To function the `nnir-hyrex-remove-prefix' variable must also be
+;; correct, see the documentation for `nnir-wais-remove-prefix' above.
+
+;; 4. find-grep
+;;
+;; The find-grep engine simply runs find(1) to locate eligible
+;; articles and searches them with grep(1).  This, of course, is much
+;; slower than using a proper search engine but OTOH doesn't require
+;; maintenance of an index and is still faster than using any built-in
+;; means for searching.  The method specification of the server to
+;; search must include a directory for this engine to work (E.g.,
+;; `nnml-directory').  The tools must be POSIX compliant.  GNU Find
+;; prior to version 4.2.12 (4.2.26 on Linux due to incorrect ARG_MAX
+;; handling) does not work.
+;; ,----
+;; |    ;; find-grep configuration for searching the Gnus Cache
+;; |
+;; |	(nnml "cache"
+;; |          (nnml-get-new-mail nil)
+;; |          (nnir-search-engine find-grep)
+;; |          (nnml-directory "~/News/cache/")
+;; |          (nnml-active-file "~/News/cache/active"))
+;; `----
+
+;; Developer information:
+
+;; I have tried to make the code expandable.  Basically, it is divided
+;; into two layers.  The upper layer is somewhat like the `nnvirtual'
+;; or `nnkiboze' backends: given a specification of what articles to
+;; show from another backend, it creates a group containing exactly
+;; those articles.  The lower layer issues a query to a search engine
+;; and produces such a specification of what articles to show from the
+;; other backend.
+
+;; The interface between the two layers consists of the single
+;; function `nnir-run-query', which just selects the appropriate
+;; function for the search engine one is using.  The input to
+;; `nnir-run-query' is a string, representing the query as input by
+;; the user.  The output of `nnir-run-query' is supposed to be a
+;; vector, each element of which should in turn be a three-element
+;; vector.  The first element should be full group name of the article,
+;; the second element should be the article number, and the third
+;; element should be the Retrieval Status Value (RSV) as returned from
+;; the search engine.  An RSV is the score assigned to the document by
+;; the search engine.  For Boolean search engines, the
+;; RSV is always 1000 (or 1 or 100, or whatever you like).
+
+;; The sorting order of the articles in the summary buffer created by
+;; nnir is based on the order of the articles in the above mentioned
+;; vector, so that's where you can do the sorting you'd like.  Maybe
+;; it would be nice to have a way of displaying the search result
+;; sorted differently?
+
+;; So what do you need to do when you want to add another search
+;; engine?  You write a function that executes the query.  Temporary
+;; data from the search engine can be put in `nnir-tmp-buffer'.  This
+;; function should return the list of articles as a vector, as
+;; described above.  Then, you need to register this backend in
+;; `nnir-engines'.  Then, users can choose the backend by setting
+;; `nnir-search-engine'.
+
+;; Todo, or future ideas:
+
+;; * It should be possible to restrict search to certain groups.
+;;
+;; * There is currently no error checking.
+;;
+;; * The summary buffer display is currently really ugly, with all the
+;;   added information in the subjects.  How could I make this
+;;   prettier?
+;;
+;; * A function which can be called from an nnir summary buffer which
+;;   teleports you into the group the current article came from and
+;;   shows you the whole thread this article is part of.
+;;   Implementation suggestions?
+;;   (1998-07-24: There is now a preliminary implementation, but
+;;   it is much too slow and quite fragile.)
+;;
+;; * Support other mail backends.  In particular, probably quite a few
+;;   people use nnfolder.  How would one go about searching nnfolders
+;;   and producing the right data needed?  The group name and the RSV
+;;   are simple, but what about the article number?
+;;   - The article number is encoded in the `X-Gnus-Article-Number'
+;;     header of each mail.
+;;   - The HyREX engine supports nnfolder.
+;;
+;; * Support compressed mail files.  Probably, just stripping off the
+;;   `.gz' or `.Z' file name extension is sufficient.
+;;
+;; * At least for imap, the query is performed twice.
+;;
+
+;; Have you got other ideas?
+
+;;; Setup Code:
+
+(require 'nnoo)
+(require 'gnus-group)
+(require 'gnus-sum)
+(require 'message)
+(require 'gnus-util)
+(eval-and-compile
+  (require 'cl))
+
+(nnoo-declare nnir)
+(nnoo-define-basics nnir)
+
+(gnus-declare-backend "nnir" 'mail)
+
+(defvar nnir-imap-search-field "TEXT"
+  "The IMAP search item when doing an nnir search")
+
+(defvar nnir-imap-search-arguments
+  '(("Whole message" . "TEXT")
+    ("Subject" . "SUBJECT")
+    ("To" . "TO")
+    ("From" . "FROM")
+    (nil . "HEADER \"%s\""))
+  "Mapping from user readable strings to IMAP search items for use in nnir")
+
+(defvar nnir-imap-search-argument-history ()
+  "The history for querying search options in nnir")
+
+;;; Developer Extension Variable:
+
+(defvar nnir-engines
+  `((wais    nnir-run-waissearch
+             ())
+    (imap    nnir-run-imap
+             ((criteria 
+	       "Search in: "                      ; Prompt
+	       ,nnir-imap-search-arguments        ; alist for completing
+	       nil                                ; no filtering
+	       nil                                ; allow any user input
+	       nil                                ; initial value
+	       nnir-imap-search-argument-history  ; the history to use
+	       ,nnir-imap-search-field            ; default
+	       )))
+    (swish++ nnir-run-swish++
+             ((group . "Group spec: ")))
+    (swish-e nnir-run-swish-e
+             ((group . "Group spec: ")))
+    (namazu  nnir-run-namazu
+             ())
+    (hyrex   nnir-run-hyrex
+	     ((group . "Group spec: ")))
+  (find-grep nnir-run-find-grep
+	     ((grep-options . "Grep options: "))))
+  "Alist of supported search engines.
+Each element in the alist is a three-element list (ENGINE FUNCTION ARGS).
+ENGINE is a symbol designating the searching engine.  FUNCTION is also
+a symbol, giving the function that does the search.  The third element
+ARGS is a list of cons pairs (PARAM . PROMPT).  When issuing a query,
+the FUNCTION will issue a query for each of the PARAMs, using PROMPT.
+
+The value of `nnir-search-engine' must be one of the ENGINE symbols.
+For example, use the following line for searching using freeWAIS-sf:
+    (setq nnir-search-engine 'wais)
+Use the following line if you read your mail via IMAP and your IMAP
+server supports searching:
+    (setq nnir-search-engine 'imap)
+Note that you have to set additional variables for most backends.  For
+example, the `wais' backend needs the variables `nnir-wais-program',
+`nnir-wais-database' and `nnir-wais-remove-prefix'.
+
+Add an entry here when adding a new search engine.")
+
+;;; User Customizable Variables:
+
+(defgroup nnir nil
+  "Search nnmh and nnml groups in Gnus with swish-e, freeWAIS-sf, or EWS."
+  :group 'gnus)
+
+;; Mail backend.
+
+;; TODO:
+;; If `nil', use server parameters to find out which server to search. CCC
+;;
+(defcustom nnir-mail-backend '(nnml "")
+  "*Specifies which backend should be searched.
+More precisely, this is used to determine from which backend to fetch the
+messages found.
+
+This must be equal to an existing server, so maybe it is best to use
+something like the following:
+    (setq nnir-mail-backend (nth 0 gnus-secondary-select-methods))
+The above line works fine if the mail backend you want to search is
+the first element of gnus-secondary-select-methods (`nth' starts counting
+at zero)."
+  :type '(sexp)
+  :group 'nnir)
+
+;; Search engine to use.
+
+(defcustom nnir-search-engine 'wais
+  "*The search engine to use.  Must be a symbol.
+See `nnir-engines' for a list of supported engines, and for example
+settings of `nnir-search-engine'."
+  :type '(sexp)
+  :group 'nnir)
+
+;; freeWAIS-sf.
+
+(defcustom nnir-wais-program "waissearch"
+  "*Name of waissearch executable."
+  :type '(string)
+  :group 'nnir)
+
+(defcustom nnir-wais-database (expand-file-name "~/.wais/mail")
+  "*Name of Wais database containing the mail.
+
+Note that this should be a file name without extension.  For example,
+if you have a file /home/john/.wais/mail.fmt, use this:
+    (setq nnir-wais-database \"/home/john/.wais/mail\")
+The string given here is passed to `waissearch -d' as-is."
+  :type '(file)
+  :group 'nnir)
+
+(defcustom nnir-wais-remove-prefix (concat (getenv "HOME") "/Mail/")
+  "*The prefix to remove from each directory name returned by waissearch
+in order to get a group name (albeit with / instead of .).  This is a
+regular expression.
+
+For example, suppose that Wais returns file names such as
+\"/home/john/Mail/mail/misc/42\".  For this example, use the following
+setting:  (setq nnir-wais-remove-prefix \"/home/john/Mail/\")
+Note the trailing slash.  Removing this prefix gives \"mail/misc/42\".
+`nnir' knows to remove the \"/42\" and to replace \"/\" with \".\" to
+arrive at the correct group name, \"mail.misc\"."
+  :type '(regexp)
+  :group 'nnir)
+
+(defcustom nnir-swish++-configuration-file
+  (expand-file-name "~/Mail/swish++.conf")
+  "*Configuration file for swish++."
+  :type '(file)
+  :group 'nnir)
+
+(defcustom nnir-swish++-program "search"
+  "*Name of swish++ search executable."
+  :type '(string)
+  :group 'nnir)
+
+(defcustom nnir-swish++-additional-switches '()
+  "*A list of strings, to be given as additional arguments to swish++.
+
+Note that this should be a list.  Ie, do NOT use the following:
+    (setq nnir-swish++-additional-switches \"-i -w\") ; wrong
+Instead, use this:
+    (setq nnir-swish++-additional-switches '(\"-i\" \"-w\"))"
+  :type '(repeat (string))
+  :group 'nnir)
+
+(defcustom nnir-swish++-remove-prefix (concat (getenv "HOME") "/Mail/")
+  "*The prefix to remove from each file name returned by swish++
+in order to get a group name (albeit with / instead of .).  This is a
+regular expression.
+
+This variable is very similar to `nnir-wais-remove-prefix', except
+that it is for swish++, not Wais."
+  :type '(regexp)
+  :group 'nnir)
+
+;; Swish-E.
+;; URL: http://sunsite.berkeley.edu/SWISH-E/
+;; New version: http://www.boe.es/swish-e
+;; Variables `nnir-swish-e-index-file', `nnir-swish-e-program' and
+;; `nnir-swish-e-additional-switches'
+
+(make-obsolete-variable 'nnir-swish-e-index-file
+			'nnir-swish-e-index-files)
+(defcustom nnir-swish-e-index-file
+  (expand-file-name "~/Mail/index.swish-e")
+  "*Index file for swish-e.
+This could be a server parameter.
+It is never consulted once `nnir-swish-e-index-files', which should be
+used instead, has been customized."
+  :type '(file)
+  :group 'nnir)
+
+(defcustom nnir-swish-e-index-files
+  (list nnir-swish-e-index-file)
+  "*List of index files for swish-e.
+This could be a server parameter."
+  :type '(repeat (file))
+  :group 'nnir)
+
+(defcustom nnir-swish-e-program "swish-e"
+  "*Name of swish-e search executable.
+This cannot be a server parameter."
+  :type '(string)
+  :group 'nnir)
+
+(defcustom nnir-swish-e-additional-switches '()
+  "*A list of strings, to be given as additional arguments to swish-e.
+
+Note that this should be a list.  Ie, do NOT use the following:
+    (setq nnir-swish-e-additional-switches \"-i -w\") ; wrong
+Instead, use this:
+    (setq nnir-swish-e-additional-switches '(\"-i\" \"-w\"))
+
+This could be a server parameter."
+  :type '(repeat (string))
+  :group 'nnir)
+
+(defcustom nnir-swish-e-remove-prefix (concat (getenv "HOME") "/Mail/")
+  "*The prefix to remove from each file name returned by swish-e
+in order to get a group name (albeit with / instead of .).  This is a
+regular expression.
+
+This variable is very similar to `nnir-wais-remove-prefix', except
+that it is for swish-e, not Wais.
+
+This could be a server parameter."
+  :type '(regexp)
+  :group 'nnir)
+
+;; HyREX engine, see <URL:http://ls6-www.cs.uni-dortmund.de/>
+
+(defcustom nnir-hyrex-program "nnir-search"
+  "*Name of the nnir-search executable."
+  :type '(string)
+  :group 'nnir)
+
+(defcustom nnir-hyrex-additional-switches '()
+  "*A list of strings, to be given as additional arguments for nnir-search.
+Note that this should be a list. Ie, do NOT use the following:
+    (setq nnir-hyrex-additional-switches \"-ddl ddl.xml -c nnir\") ; wrong !
+Instead, use this:
+    (setq nnir-hyrex-additional-switches '(\"-ddl\" \"ddl.xml\" \"-c\" \"nnir\"))"
+  :type '(repeat (string))
+  :group 'nnir)
+
+(defcustom nnir-hyrex-index-directory (getenv "HOME")
+  "*Index directory for HyREX."
+  :type '(directory)
+  :group 'nnir)
+
+(defcustom nnir-hyrex-remove-prefix (concat (getenv "HOME") "/Mail/")
+  "*The prefix to remove from each file name returned by HyREX
+in order to get a group name (albeit with / instead of .).
+
+For example, suppose that HyREX returns file names such as
+\"/home/john/Mail/mail/misc/42\".  For this example, use the following
+setting:  (setq nnir-hyrex-remove-prefix \"/home/john/Mail/\")
+Note the trailing slash.  Removing this prefix gives \"mail/misc/42\".
+`nnir' knows to remove the \"/42\" and to replace \"/\" with \".\" to
+arrive at the correct group name, \"mail.misc\"."
+  :type '(directory)
+  :group 'nnir)
+
+;; Namazu engine, see <URL:http://ww.namazu.org/>
+
+(defcustom nnir-namazu-program "namazu"
+  "*Name of Namazu search executable."
+  :type '(string)
+  :group 'nnir)
+
+(defcustom nnir-namazu-index-directory (expand-file-name "~/Mail/namazu/")
+  "*Index directory for Namazu."
+  :type '(directory)
+  :group 'nnir)
+
+(defcustom nnir-namazu-additional-switches '()
+  "*A list of strings, to be given as additional arguments to namazu.
+The switches `-q', `-a', and `-s' are always used, very few other switches
+make any sense in this context.
+
+Note that this should be a list.  Ie, do NOT use the following:
+    (setq nnir-namazu-additional-switches \"-i -w\") ; wrong
+Instead, use this:
+    (setq nnir-namazu-additional-switches '(\"-i\" \"-w\"))"
+  :type '(repeat (string))
+  :group 'nnir)
+
+(defcustom nnir-namazu-remove-prefix (concat (getenv "HOME") "/Mail/")
+  "*The prefix to remove from each file name returned by Namazu
+in order to get a group name (albeit with / instead of .).
+
+This variable is very similar to `nnir-wais-remove-prefix', except
+that it is for Namazu, not Wais."
+  :type '(directory)
+  :group 'nnir)
+
+;;; Internal Variables:
+
+(defvar nnir-current-query nil
+  "Internal: stores current query (= group name).")
+
+(defvar nnir-current-server nil
+  "Internal: stores current server (does it ever change?).")
+
+(defvar nnir-current-group-marked nil
+  "Internal: stores current list of process-marked groups.")
+
+(defvar nnir-artlist nil
+  "Internal: stores search result.")
+
+(defvar nnir-tmp-buffer " *nnir*"
+  "Internal: temporary buffer.")
+
+;;; Code:
+
+;; Gnus glue.
+
+(defun gnus-group-make-nnir-group (extra-parms query)
+  "Create an nnir group.  Asks for query."
+  (interactive "P\nsQuery: ")
+  (setq nnir-current-query nil
+	nnir-current-server nil
+	nnir-current-group-marked nil
+	nnir-artlist nil)
+  (let ((parms nil))
+    (if extra-parms
+        (setq parms (nnir-read-parms query))
+      (setq parms (list (cons 'query query))))
+    (add-to-list 'parms (cons 'unique-id (message-unique-id)) t)
+    (gnus-group-read-ephemeral-group
+     (concat "nnir:" (prin1-to-string parms)) '(nnir "") t
+     (cons (current-buffer)
+           gnus-current-window-configuration)
+     nil)))
+
+(defun nnir-group-mode-hook ()
+  (define-key gnus-group-mode-map (kbd "G G")
+    'gnus-group-make-nnir-group))
+(add-hook 'gnus-group-mode-hook 'nnir-group-mode-hook)
+
+;; Why is this needed? Is this for compatibility with old/new gnusae? Using
+;; gnus-group-server instead works for me.  -- Justus Piater
+(defmacro nnir-group-server (group)
+  "Return the server for a newsgroup GROUP.
+The returned format is as `gnus-server-to-method' needs it.  See
+`gnus-group-real-prefix' and `gnus-group-real-name'."
+  `(let ((gname ,group))
+     (if (string-match "^\\([^:]+\\):" gname)
+	 (progn
+	   (setq gname (match-string 1 gname))
+	   (if (string-match "^\\([^+]+\\)\\+\\(.+\\)$" gname)
+	       (format "%s:%s" (match-string 1 gname) (match-string 2 gname))
+	     (concat gname ":")))
+       (format "%s:%s" (car gnus-select-method) (cadr gnus-select-method)))))
+
+;; Summary mode commands.
+
+(defun gnus-summary-nnir-goto-thread ()
+  "Only applies to nnir groups.  Go to group this article came from
+and show thread that contains this article."
+  (interactive)
+  (unless (eq 'nnir (car (gnus-find-method-for-group gnus-newsgroup-name)))
+    (error "Can't execute this command unless in nnir group."))
+  (let* ((cur (gnus-summary-article-number))
+         (group (nnir-artlist-artitem-group nnir-artlist cur))
+         (backend-number (nnir-artlist-artitem-number nnir-artlist cur))
+	 server backend-group)
+    (setq server (nnir-group-server group))
+    (setq backend-group (gnus-group-real-name group))
+    (gnus-group-read-ephemeral-group
+     backend-group
+     (gnus-server-to-method server)
+     t                                  ; activate
+     (cons (current-buffer)
+           'summary)                    ; window config
+     nil
+     (list backend-number))
+    (gnus-summary-limit (list backend-number))
+    (gnus-summary-refer-thread)))
+
+(if (fboundp 'eval-after-load)
+    (eval-after-load "gnus-sum"
+      '(define-key gnus-summary-goto-map
+         "T" 'gnus-summary-nnir-goto-thread))
+  (add-hook 'gnus-summary-mode-hook
+            (function (lambda ()
+                        (define-key gnus-summary-goto-map
+                          "T" 'gnus-summary-nnir-goto-thread)))))
+
+
+
+;; Gnus backend interface functions.
+
+(deffoo nnir-open-server (server &optional definitions)
+  ;; Just set the server variables appropriately.
+  (nnoo-change-server 'nnir server definitions))
+
+(deffoo nnir-request-group (group &optional server fast)
+  "GROUP is the query string."
+  (nnir-possibly-change-server server)
+  ;; Check for cache and return that if appropriate.
+  (if (and (equal group nnir-current-query)
+           (equal gnus-group-marked nnir-current-group-marked)
+           (or (null server)
+               (equal server nnir-current-server)))
+      nnir-artlist
+    ;; Cache miss.
+    (setq nnir-artlist (nnir-run-query group)))
+  (save-excursion
+    (set-buffer nntp-server-buffer)
+    (if (zerop (length nnir-artlist))
+	(progn
+	  (setq nnir-current-query nil
+		nnir-current-server nil
+		nnir-current-group-marked nil
+		nnir-artlist nil)
+	  (nnheader-report 'nnir "Search produced empty results."))
+      ;; Remember data for cache.
+      (setq nnir-current-query group)
+      (when server (setq nnir-current-server server))
+      (setq nnir-current-group-marked gnus-group-marked)
+      (nnheader-insert "211 %d %d %d %s\n"
+		       (nnir-artlist-length nnir-artlist) ; total #
+		       1              ; first #
+		       (nnir-artlist-length nnir-artlist) ; last #
+		       group))))     ; group name
+
+(deffoo nnir-retrieve-headers (articles &optional group server fetch-old)
+  (save-excursion
+    (let ((artlist (copy-sequence articles))
+          art artitem artgroup artno artrsv artfullgroup
+          novitem novdata foo server)
+      (while (not (null artlist))
+        (setq art (car artlist))
+        (or (numberp art)
+            (nnheader-report
+             'nnir
+             "nnir-retrieve-headers doesn't grok message ids: %s"
+             art))
+        (setq artitem (nnir-artlist-article nnir-artlist art))
+        (setq artrsv (nnir-artitem-rsv artitem))
+        (setq artfullgroup (nnir-artitem-group artitem))
+        (setq artno (nnir-artitem-number artitem))
+        (setq artgroup (gnus-group-real-name artfullgroup))
+	(setq server (nnir-group-server artfullgroup))
+        ;; retrieve NOV or HEAD data for this article, transform into
+        ;; NOV data and prepend to `novdata'
+        (set-buffer nntp-server-buffer)
+	(nnir-possibly-change-server server)
+        (let ((gnus-override-method
+	       (gnus-server-to-method server)))
+	  (case (setq foo (gnus-retrieve-headers (list artno) artfullgroup nil))
+	    (nov
+	     (goto-char (point-min))
+	     (setq novitem (nnheader-parse-nov))
+	     (unless novitem
+	       (pop-to-buffer nntp-server-buffer)
+	       (error
+		"nnheader-parse-nov returned nil for article %s in group %s"
+		artno artfullgroup)))
+	    (headers
+	     (goto-char (point-min))
+	     (setq novitem (nnheader-parse-head))
+	     (unless novitem
+	       (pop-to-buffer nntp-server-buffer)
+	       (error
+		"nnheader-parse-head returned nil for article %s in group %s"
+		artno artfullgroup)))
+	    (t (error "Unknown header type %s while requesting article %s of group %s"
+		      foo artno artfullgroup))))
+	;; replace article number in original group with article number
+        ;; in nnir group
+        (mail-header-set-number novitem art)
+        (mail-header-set-from novitem
+                              (mail-header-from novitem))
+        (mail-header-set-subject
+         novitem
+         (format "[%d: %s/%d] %s"
+                 artrsv artgroup artno
+                 (mail-header-subject novitem)))
+        ;;-(mail-header-set-extra novitem nil)
+        (push novitem novdata)
+        (setq artlist (cdr artlist)))
+      (setq novdata (nreverse novdata))
+      (set-buffer nntp-server-buffer) (erase-buffer)
+      (mapc 'nnheader-insert-nov novdata)
+      'nov)))
+
+(deffoo nnir-request-article (article
+                              &optional group server to-buffer)
+  (if (stringp article)
+      (nnheader-report
+       'nnir
+       "nnir-retrieve-headers doesn't grok message ids: %s"
+       article)
+    (save-excursion
+      (let* ((artitem (nnir-artlist-article nnir-artlist
+					    article))
+	     (artfullgroup (nnir-artitem-group artitem))
+	     (artno (nnir-artitem-number artitem))
+	     ;; Bug?
+	     ;; Why must we bind nntp-server-buffer here?  It won't
+	     ;; work if `buf' is used, say.  (Of course, the set-buffer
+	     ;; line below must then be updated, too.)
+	     (nntp-server-buffer (or to-buffer nntp-server-buffer)))
+	(set-buffer nntp-server-buffer)
+	(erase-buffer)
+	(message "Requesting article %d from group %s"
+		 artno artfullgroup)
+	(gnus-request-article artno artfullgroup nntp-server-buffer)
+	(cons artfullgroup artno)))))
+
+
+(nnoo-define-skeleton nnir)
+
+
+(defmacro nnir-add-result (dirnam artno score prefix server artlist)
+  "Ask `nnir-compose-result' to construct a result vector, 
+and if it is non-nil, add it to artlist."
+  `(let ((result (nnir-compose-result ,dirnam ,artno ,score ,prefix ,server)))
+     (when (not (null result))
+       (push result ,artlist))))
+
+(autoload 'nnmaildir-base-name-to-article-number "nnmaildir")
+
+;; Helper function currently used by the Swish++ and Namazu backends;
+;; perhaps useful for other backends as well
+(defun nnir-compose-result (dirnam article score prefix server)
+  "Extract the group from dirnam, and create a result vector
+ready to be added to the list of search results."
+
+  ;; remove nnir-*-remove-prefix from beginning of dirnam filename
+  (when (string-match (concat "^" prefix) dirnam)
+    (setq dirnam (replace-match "" t t dirnam)))
+
+  (when (file-readable-p (concat prefix dirnam article))
+    ;; remove trailing slash and, for nnmaildir, cur/new/tmp
+    (setq dirnam
+	  (substring dirnam 0 (if (string= server "nnmaildir:") -5 -1)))
+
+    ;; Set group to dirnam without any leading dots or slashes,
+    ;; and with all subsequent slashes replaced by dots
+    (let ((group (gnus-replace-in-string
+                 (gnus-replace-in-string dirnam "^[./\\]" "" t)
+                 "[/\\]" "." t)))
+
+    (vector (nnir-group-full-name group server)
+	    (if (string= server "nnmaildir:")
+		(nnmaildir-base-name-to-article-number
+		 (substring article 0 (string-match ":" article))
+		 group nil)
+	      (string-to-number article))
+	    (string-to-number score)))))
+
+;;; Search Engine Interfaces:
+
+;; freeWAIS-sf interface.
+(defun nnir-run-waissearch (query server &optional group)
+  "Run given query agains waissearch.  Returns vector of (group name, file name)
+pairs (also vectors, actually)."
+  (when group
+    (error "The freeWAIS-sf backend cannot search specific groups."))
+  (save-excursion
+    (let ((qstring (cdr (assq 'query query)))
+	  (prefix (nnir-read-server-parm 'nnir-wais-remove-prefix server))
+          artlist score artno dirnam)
+      (set-buffer (get-buffer-create nnir-tmp-buffer))
+      (erase-buffer)
+      (message "Doing WAIS query %s..." query)
+      (call-process nnir-wais-program
+                    nil                 ; input from /dev/null
+                    t                   ; output to current buffer
+                    nil                 ; don't redisplay
+                    "-d" (nnir-read-server-parm 'nnir-wais-database server) ; database to search
+                    qstring)
+      (message "Massaging waissearch output...")
+      ;; remove superfluous lines
+      (keep-lines "Score:")
+      ;; extract data from result lines
+      (goto-char (point-min))
+      (while (re-search-forward
+              "Score: +\\([0-9]+\\).*'\\([0-9]+\\) +\\([^']+\\)/'" nil t)
+        (setq score (match-string 1)
+              artno (match-string 2)
+              dirnam (match-string 3))
+        (unless (string-match prefix dirnam)
+          (nnheader-report 'nnir "Dir name %s doesn't contain prefix %s"
+                           dirnam prefix))
+        (setq group (substitute ?. ?/ (replace-match "" t t dirnam)))
+        (push (vector (nnir-group-full-name group server)
+                      (string-to-number artno)
+                      (string-to-number score))
+              artlist))
+      (message "Massaging waissearch output...done")
+      (apply 'vector
+             (sort* artlist
+                    (function (lambda (x y)
+                                (> (nnir-artitem-rsv x)
+                                   (nnir-artitem-rsv y)))))))))
+
+;; IMAP interface.
+;; todo:
+;; nnir invokes this two (2) times???!
+;; we should not use nnimap at all but open our own server connection
+;; we should not LIST * but use nnimap-list-pattern from defs
+;; send queries as literals
+;; handle errors
+
+(autoload 'nnimap-open-server "nnimap")
+(defvar nnimap-server-buffer) ;; nnimap.el
+(autoload 'imap-mailbox-select "imap")
+(autoload 'imap-search "imap")
+(autoload 'imap-quote-specials "imap")
+
+(defun nnir-run-imap (query srv &optional group-option)
+  "Run a search against an IMAP back-end server.
+This uses a custom query language parser; see `nnir-imap-make-query' for
+details on the language and supported extensions"
+  (save-excursion
+    (let ((qstring (cdr (assq 'query query)))
+	  (server (cadr (gnus-server-to-method srv)))
+	  (group (or group-option (gnus-group-group-name)))
+	  (defs (caddr (gnus-server-to-method srv)))
+	  (criteria (or (cdr (assq 'criteria query))
+			nnir-imap-search-field))
+	  artlist buf)
+      (message "Opening server %s" server)
+      (condition-case ()
+	  (when (nnimap-open-server server defs) ;; xxx
+	    (setq buf nnimap-server-buffer) ;; xxx
+	    (message "Searching %s..." group)
+            (let ((arts 0)
+                  (mbx (gnus-group-real-name group)))
+              (when (imap-mailbox-select mbx nil buf)
+                (mapc
+                 (lambda (artnum)
+                   (push (vector group artnum 1) artlist)
+                   (setq arts (1+ arts)))
+                 (imap-search (nnir-imap-make-query criteria qstring) buf))
+                (message "Searching %s... %d matches" mbx arts)))
+            (message "Searching %s...done" group))
+        (quit nil))
+      (reverse artlist))))
+
+(defun nnir-imap-make-query (criteria qstring)
+  "Parse the query string and criteria into an appropriate IMAP search
+expression, returning the string query to make.
+
+This implements a little language designed to return the expected results
+to an arbitrary query string to the end user.
+
+The search is always case-insensitive, as defined by RFC2060, and supports
+the following features (inspired by the Google search input language): 
+
+Automatic \"and\" queries
+    If you specify multiple words then they will be treated as an \"and\"
+    expression intended to match all components.
+
+Phrase searches
+    If you wrap your query in double-quotes then it will be treated as a
+    literal string.
+
+Negative terms
+    If you precede a term with \"-\" then it will negate that.
+
+\"OR\" queries
+    If you include an upper-case \"OR\" in your search it will cause the
+    term before it and the term after it to be treated as alternatives.
+
+In future the following will be added to the language:
+ * support for date matches
+ * support for location of text matching within the query
+ * from/to/etc headers
+ * additional search terms
+ * flag based searching
+ * anything else that the RFC supports, basically."
+  ;; Walk through the query and turn it into an IMAP query string.
+  (nnir-imap-query-to-imap criteria (nnir-imap-parse-query qstring)))
+
+
+(defun nnir-imap-query-to-imap (criteria query)
+  "Turn a s-expression format query into IMAP."
+  (mapconcat
+   ;; Turn the expressions into IMAP text
+   (lambda (item)
+     (nnir-imap-expr-to-imap criteria item))
+   ;; The query, already in s-expr format.
+   query
+   ;; Append a space between each expression
+   " "))
+
+
+(defun nnir-imap-expr-to-imap (criteria expr)
+  "Convert EXPR into an IMAP search expression on CRITERIA"
+  ;; What sort of expression is this, eh?
+  (cond
+   ;; Simple string term
+   ((stringp expr)
+    (format "%s \"%s\"" criteria (imap-quote-specials expr)))
+   ;; Trivial term: and
+   ((eq expr 'and) nil)
+   ;; Composite term: or expression
+   ((eq (car-safe expr) 'or)
+    (format "OR %s %s"
+	    (nnir-imap-expr-to-imap criteria (second expr))
+	    (nnir-imap-expr-to-imap criteria (third expr))))
+   ;; Composite term: just the fax, mam
+   ((eq (car-safe expr) 'not)
+    (format "NOT (%s)" (nnir-imap-query-to-imap criteria (rest expr))))
+   ;; Composite term: just expand it all.
+   ((and (not (null expr)) (listp expr))
+    (format "(%s)" (nnir-imap-query-to-imap criteria expr)))
+   ;; Complex value, give up for now.
+   (t (error "Unhandled input: %S" expr))))
+
+
+(defun nnir-imap-parse-query (string)
+  "Turn STRING into an s-expression based query based on the IMAP
+query language as defined in `nnir-imap-make-query'.
+
+This involves turning individual tokens into higher level terms
+that the search language can then understand and use."
+  (with-temp-buffer
+    ;; Set up the parsing environment.
+    (insert string)
+    (goto-char (point-min))
+    ;; Now, collect the output terms and return them.
+    (let (out)
+      (while (not (nnir-imap-end-of-input))
+	(push (nnir-imap-next-expr) out))
+      (reverse out))))
+
+
+(defun nnir-imap-next-expr (&optional count)
+  "Return the next expression from the current buffer."
+  (let ((term (nnir-imap-next-term count))
+	(next (nnir-imap-peek-symbol)))
+    ;; Are we looking at an 'or' expression?
+    (cond
+     ;; Handle 'expr or expr'
+     ((eq next 'or)
+      (list 'or term (nnir-imap-next-expr 2)))
+     ;; Anything else
+     (t term))))
+
+
+(defun nnir-imap-next-term (&optional count)
+  "Return the next TERM from the current buffer."
+  (let ((term (nnir-imap-next-symbol count)))
+    ;; What sort of term is this?
+    (cond
+     ;; and -- just ignore it
+     ((eq term 'and) 'and)
+     ;; negated term
+     ((eq term 'not) (list 'not (nnir-imap-next-expr)))
+     ;; generic term
+     (t term))))
+
+
+(defun nnir-imap-peek-symbol ()
+  "Return the next symbol from the current buffer, but don't consume it."
+  (save-excursion
+    (nnir-imap-next-symbol)))
+
+(defun nnir-imap-next-symbol (&optional count)
+  "Return the next symbol from the current buffer, or nil if we are
+at the end of the buffer.  If supplied COUNT skips some symbols before
+returning the one at the supplied position."
+  (when (and (numberp count) (> count 1))
+    (nnir-imap-next-symbol (1- count)))
+  (let ((case-fold-search t))
+    ;; end of input stream?
+    (unless (nnir-imap-end-of-input)
+      ;; No, return the next symbol from the stream.
+      (cond
+       ;; negated expression -- return it and advance one char.
+       ((looking-at "-") (forward-char 1) 'not)
+       ;; quoted string
+       ((looking-at "\"") (nnir-imap-delimited-string "\""))
+       ;; list expression -- we parse the content and return this as a list.
+       ((looking-at "(")
+	(nnir-imap-parse-query (nnir-imap-delimited-string ")")))
+       ;; keyword input -- return a symbol version
+       ((looking-at "\\band\\b") (forward-char 3) 'and)
+       ((looking-at "\\bor\\b")  (forward-char 2) 'or)
+       ((looking-at "\\bnot\\b") (forward-char 3) 'not)
+       ;; Simple, boring keyword
+       (t (let ((start (point))
+		(end (if (search-forward-regexp "[[:blank:]]" nil t)
+			 (prog1
+			     (match-beginning 0)
+			   ;; unskip if we hit a non-blank terminal character.
+			   (when (string-match "[^[:blank:]]" (match-string 0))
+			     (backward-char 1)))
+		       (goto-char (point-max)))))
+	    (buffer-substring start end)))))))
+
+(defun nnir-imap-delimited-string (delimiter)
+  "Return a delimited string from the current buffer."
+  (let ((start (point)) end)
+    (forward-char 1)			; skip the first delimiter.
+    (while (not end)
+      (unless (search-forward delimiter nil t)
+	(error "Unmatched delimited input with %s in query" delimiter))
+      (let ((here (point)))
+	(unless (equal (buffer-substring (- here 2) (- here 1)) "\\")
+	  (setq end (point)))))
+    (buffer-substring (1+ start) (1- end))))
+
+(defun nnir-imap-end-of-input ()
+  "Are we at the end of input?"
+  (skip-chars-forward "[[:blank:]]")
+  (looking-at "$"))
+  
+
+;; Swish++ interface.
+;; -cc- Todo
+;; Search by
+;; - group
+;; Sort by
+;; - rank (default)
+;; - article number
+;; - file size
+;; - group
+(defun nnir-run-swish++ (query server &optional group)
+  "Run QUERY against swish++.
+Returns a vector of (group name, file name) pairs (also vectors,
+actually).
+
+Tested with swish++ 4.7 on GNU/Linux and with swish++ 5.0b2 on
+Windows NT 4.0."
+
+  (when group
+    (error "The swish++ backend cannot search specific groups."))
+
+  (save-excursion
+    (let ( (qstring (cdr (assq 'query query)))
+	   (groupspec (cdr (assq 'group query)))
+	   (prefix (nnir-read-server-parm 'nnir-swish++-remove-prefix server))
+           artlist
+	   ;; nnml-use-compressed-files might be any string, but probably this
+	   ;; is sufficient.  Note that we can't only use the value of
+	   ;; nnml-use-compressed-files because old articles might have been
+	   ;; saved with a different value.
+	   (article-pattern (if (string= server "nnmaildir:")
+				":[0-9]+"
+			      "^[0-9]+\\(\\.[a-z0-9]+\\)?$"))
+           score artno dirnam filenam)
+
+      (when (equal "" qstring)
+        (error "swish++: You didn't enter anything."))
+
+      (set-buffer (get-buffer-create nnir-tmp-buffer))
+      (erase-buffer)
+
+      (if groupspec
+          (message "Doing swish++ query %s on %s..." qstring groupspec)
+        (message "Doing swish++ query %s..." qstring))
+
+      (let* ((cp-list `( ,nnir-swish++-program
+                         nil            ; input from /dev/null
+                         t              ; output
+                         nil            ; don't redisplay
+                         "--config-file" ,(nnir-read-server-parm 'nnir-swish++-configuration-file server)
+                         ,@(nnir-read-server-parm 'nnir-swish++-additional-switches server)
+                         ,qstring       ; the query, in swish++ format
+                         ))
+             (exitstatus
+              (progn
+                (message "%s args: %s" nnir-swish++-program
+                         (mapconcat 'identity (cddddr cp-list) " ")) ;; ???
+                (apply 'call-process cp-list))))
+        (unless (or (null exitstatus)
+                    (zerop exitstatus))
+          (nnheader-report 'nnir "Couldn't run swish++: %s" exitstatus)
+          ;; swish++ failure reason is in this buffer, show it if
+          ;; the user wants it.
+          (when (> gnus-verbose 6)
+            (display-buffer nnir-tmp-buffer))))
+
+      ;; The results are output in the format of:
+      ;; V 4.7 Linux
+      ;; rank relative-path-name file-size file-title
+      ;; V 5.0b2:
+      ;; rank relative-path-name file-size topic??
+      ;; where rank is an integer from 1 to 100.
+      (goto-char (point-min))
+      (while (re-search-forward
+              "\\(^[0-9]+\\) \\([^ ]+\\) [0-9]+ \\(.*\\)$" nil t)
+        (setq score (match-string 1)
+	      filenam (match-string 2)
+              artno (file-name-nondirectory filenam)
+              dirnam (file-name-directory filenam))
+
+        ;; don't match directories
+        (when (string-match article-pattern artno)
+          (when (not (null dirnam))
+
+	    ;; maybe limit results to matching groups.
+	    (when (or (not groupspec)
+		      (string-match groupspec dirnam))
+	      (nnir-add-result dirnam artno score prefix server artlist)))))
+
+      (message "Massaging swish++ output...done")
+
+      ;; Sort by score
+      (apply 'vector
+             (sort* artlist
+                    (function (lambda (x y)
+                                (> (nnir-artitem-rsv x)
+                                   (nnir-artitem-rsv y)))))))))
+
+;; Swish-E interface.
+(defun nnir-run-swish-e (query server &optional group)
+  "Run given query against swish-e.
+Returns a vector of (group name, file name) pairs (also vectors,
+actually).
+
+Tested with swish-e-2.0.1 on Windows NT 4.0."
+
+  ;; swish-e crashes with empty parameter to "-w" on commandline...
+  (when group
+    (error "The swish-e backend cannot search specific groups."))
+
+  (save-excursion
+    (let ((qstring (cdr (assq 'query query)))
+	  (prefix
+	   (or (nnir-read-server-parm 'nnir-swish-e-remove-prefix server)
+	       (error "Missing parameter `nnir-swish-e-remove-prefix'")))
+          artlist score artno dirnam group )
+
+      (when (equal "" qstring)
+        (error "swish-e: You didn't enter anything."))
+
+      (set-buffer (get-buffer-create nnir-tmp-buffer))
+      (erase-buffer)
+
+      (message "Doing swish-e query %s..." query)
+      (let* ((index-files
+	      (or (nnir-read-server-parm
+		   'nnir-swish-e-index-files server)
+		  (error "Missing parameter `nnir-swish-e-index-files'")))
+	     (additional-switches
+	      (nnir-read-server-parm
+	       'nnir-swish-e-additional-switches server))
+	     (cp-list `(,nnir-swish-e-program
+			nil		; input from /dev/null
+			t		; output
+			nil		; don't redisplay
+			"-f" ,@index-files
+			,@additional-switches
+			"-w"
+			,qstring	; the query, in swish-e format
+			))
+             (exitstatus
+              (progn
+                (message "%s args: %s" nnir-swish-e-program
+                         (mapconcat 'identity (cddddr cp-list) " "))
+                (apply 'call-process cp-list))))
+        (unless (or (null exitstatus)
+                    (zerop exitstatus))
+          (nnheader-report 'nnir "Couldn't run swish-e: %s" exitstatus)
+          ;; swish-e failure reason is in this buffer, show it if
+          ;; the user wants it.
+          (when (> gnus-verbose 6)
+            (display-buffer nnir-tmp-buffer))))
+
+      ;; The results are output in the format of:
+      ;; rank path-name file-title file-size
+      (goto-char (point-min))
+      (while (re-search-forward
+              "\\(^[0-9]+\\) \\([^ ]+\\) \"\\([^\"]+\\)\" [0-9]+$" nil t)
+        (setq score (match-string 1)
+              artno (match-string 3)
+              dirnam (file-name-directory (match-string 2)))
+
+        ;; don't match directories
+        (when (string-match "^[0-9]+$" artno)
+          (when (not (null dirnam))
+
+	    ;; remove nnir-swish-e-remove-prefix from beginning of dirname
+            (when (string-match (concat "^" prefix) dirnam)
+              (setq dirnam (replace-match "" t t dirnam)))
+
+            (setq dirnam (substring dirnam 0 -1))
+	    ;; eliminate all ".", "/", "\" from beginning. Always matches.
+            (string-match "^[./\\]*\\(.*\\)$" dirnam)
+            ;; "/" -> "."
+            (setq group (substitute ?. ?/ (match-string 1 dirnam)))
+            ;; Windows "\\" -> "."
+            (setq group (substitute ?. ?\\ group))
+
+            (push (vector (nnir-group-full-name group server)
+                          (string-to-number artno)
+                          (string-to-number score))
+                  artlist))))
+
+      (message "Massaging swish-e output...done")
+
+      ;; Sort by score
+      (apply 'vector
+             (sort* artlist
+                    (function (lambda (x y)
+                                (> (nnir-artitem-rsv x)
+                                   (nnir-artitem-rsv y)))))))))
+
+;; HyREX interface
+(defun nnir-run-hyrex (query server &optional group)
+  (save-excursion
+    (let ((artlist nil)
+          (groupspec (cdr (assq 'group query)))
+          (qstring (cdr (assq 'query query)))
+	  (prefix (nnir-read-server-parm 'nnir-hyrex-remove-prefix server))
+	  score artno dirnam)
+      (when (and group groupspec)
+        (error (concat "It does not make sense to use a group spec"
+                       " with process-marked groups.")))
+      (when group
+        (setq groupspec (gnus-group-real-name group)))
+      (when (and group (not (equal group (nnir-group-full-name groupspec server))))
+	(message "%s vs. %s" group (nnir-group-full-name groupspec server))
+	(error "Server with groupspec doesn't match group !"))
+      (set-buffer (get-buffer-create nnir-tmp-buffer))
+      (erase-buffer)
+      (if groupspec
+          (message "Doing hyrex-search query %s on %s..." query groupspec)
+        (message "Doing hyrex-search query %s..." query))
+      (let* ((cp-list
+	      `( ,nnir-hyrex-program
+		 nil			; input from /dev/null
+		 t			; output
+		 nil			; don't redisplay
+		 "-i",(nnir-read-server-parm 'nnir-hyrex-index-directory server) ; index directory
+		 ,@(nnir-read-server-parm 'nnir-hyrex-additional-switches server)
+		 ,qstring	   ; the query, in hyrex-search format
+		 ))
+             (exitstatus
+              (progn
+                (message "%s args: %s" nnir-hyrex-program
+                         (mapconcat 'identity (cddddr cp-list) " "))
+                (apply 'call-process cp-list))))
+        (unless (or (null exitstatus)
+                    (zerop exitstatus))
+          (nnheader-report 'nnir "Couldn't run hyrex-search: %s" exitstatus)
+          ;; nnir-search failure reason is in this buffer, show it if
+          ;; the user wants it.
+          (when (> gnus-verbose 6)
+            (display-buffer nnir-tmp-buffer)))) ;; FIXME: Dont clear buffer !
+      (if groupspec
+          (message "Doing hyrex-search query \"%s\" on %s...done" qstring groupspec)
+        (message "Doing hyrex-search query \"%s\"...done" qstring))
+      (sit-for 0)
+      ;; nnir-search returns:
+      ;;   for nnml/nnfolder: "filename mailid weigth"
+      ;;   for nnimap:        "group mailid weigth"
+      (goto-char (point-min))
+      (delete-non-matching-lines "^\\S + [0-9]+ [0-9]+$")
+      ;; HyREX couldn't search directly in groups -- so filter out here.
+      (when groupspec
+	(keep-lines groupspec))
+      ;; extract data from result lines
+      (goto-char (point-min))
+      (while (re-search-forward
+	      "\\(\\S +\\) \\([0-9]+\\) \\([0-9]+\\)" nil t)
+	(setq dirnam (match-string 1)
+	      artno (match-string 2)
+	      score (match-string 3))
+	(when (string-match prefix dirnam)
+	  (setq dirnam (replace-match "" t t dirnam)))
+	(push (vector (nnir-group-full-name (substitute ?. ?/ dirnam) server)
+		      (string-to-number artno)
+		      (string-to-number score))
+	      artlist))
+      (message "Massaging hyrex-search output...done.")
+      (apply 'vector
+	     (sort* artlist
+		    (function (lambda (x y)
+				(if (string-lessp (nnir-artitem-group x)
+						  (nnir-artitem-group y))
+				    t
+				  (< (nnir-artitem-number x)
+				     (nnir-artitem-number y)))))))
+      )))
+
+;; Namazu interface
+(defun nnir-run-namazu (query server &optional group)
+  "Run given query against Namazu.  Returns a vector of (group name, file name)
+pairs (also vectors, actually).
+
+Tested with Namazu 2.0.6 on a GNU/Linux system."
+  (when group
+    (error "The Namazu backend cannot search specific groups"))
+  (save-excursion
+    (let ((article-pattern (if (string= server "nnmaildir:")
+			       ":[0-9]+"
+			     "^[0-9]+$"))
+          artlist
+	  (qstring (cdr (assq 'query query)))
+	  (prefix (nnir-read-server-parm 'nnir-namazu-remove-prefix server))
+          score group article
+          (process-environment (copy-sequence process-environment)))
+      (setenv "LC_MESSAGES" "C")
+      (set-buffer (get-buffer-create nnir-tmp-buffer))
+      (erase-buffer)
+      (let* ((cp-list
+              `( ,nnir-namazu-program
+                 nil			; input from /dev/null
+                 t			; output
+                 nil			; don't redisplay
+                 "-q"			; don't be verbose
+                 "-a"			; show all matches
+                 "-s"			; use short format
+                 ,@(nnir-read-server-parm 'nnir-namazu-additional-switches server)
+                 ,qstring		; the query, in namazu format
+                 ,(nnir-read-server-parm 'nnir-namazu-index-directory server) ; index directory
+                 ))
+             (exitstatus
+              (progn
+                (message "%s args: %s" nnir-namazu-program
+                         (mapconcat 'identity (cddddr cp-list) " "))
+                (apply 'call-process cp-list))))
+        (unless (or (null exitstatus)
+                    (zerop exitstatus))
+          (nnheader-report 'nnir "Couldn't run namazu: %s" exitstatus)
+          ;; Namazu failure reason is in this buffer, show it if
+          ;; the user wants it.
+          (when (> gnus-verbose 6)
+            (display-buffer nnir-tmp-buffer))))
+
+      ;; Namazu output looks something like this:
+      ;; 2. Re: Gnus agent expire broken (score: 55)
+      ;; /home/henrik/Mail/mail/sent/1310 (4,138 bytes)
+
+      (goto-char (point-min))
+      (while (re-search-forward
+              "^\\([0-9]+\\.\\).*\\((score: \\([0-9]+\\)\\))\n\\([^ ]+\\)"
+              nil t)
+        (setq score (match-string 3)
+              group (file-name-directory (match-string 4))
+              article (file-name-nondirectory (match-string 4)))
+
+        ;; make sure article and group is sane
+        (when (and (string-match article-pattern article)
+                   (not (null group)))
+	  (nnir-add-result group article score prefix server artlist)))
+
+      ;; sort artlist by score
+      (apply 'vector
+             (sort* artlist
+                    (function (lambda (x y)
+                                (> (nnir-artitem-rsv x)
+                                   (nnir-artitem-rsv y)))))))))
+
+(defun nnir-run-find-grep (query server &optional group)
+  "Run find and grep to obtain matching articles."
+  (let* ((method (gnus-server-to-method server))
+	 (sym (intern
+	       (concat (symbol-name (car method)) "-directory")))
+	 (directory (cadr (assoc sym (cddr method))))
+	 (regexp (cdr (assoc 'query query)))
+	 (grep-options (cdr (assoc 'grep-options query)))
+	 artlist)
+    (unless directory
+      (error "No directory found in method specification of server %s"
+	     server))
+    (message "Searching %s using find-grep..." (or group server))
+    (save-window-excursion
+      (set-buffer (get-buffer-create nnir-tmp-buffer))
+      (erase-buffer)
+      (if (> gnus-verbose 6)
+	  (pop-to-buffer (current-buffer)))
+      (cd directory) ; Using relative paths simplifies postprocessing.
+      (let ((group
+	     (if (not group)
+		 "."
+	       ;; Try accessing the group literally as well as
+	       ;; interpreting dots as directory separators so the
+	       ;; engine works with plain nnml as well as the Gnus
+	       ;; Cache.
+	       (find-if 'file-directory-p
+		(let ((group (gnus-group-real-name group)))
+		  (list group (gnus-replace-in-string group "\\." "/" t)))))))
+	(unless group
+	  (error "Cannot locate directory for group"))
+	(save-excursion
+	  (apply
+	   'call-process "find" nil t
+	   "find" group "-type" "f" "-name" "[0-9]*" "-exec"
+	   "grep"
+	   `("-l" ,@(and grep-options (split-string grep-options "\\s-" t))
+	     "-e" ,regexp "{}" "+"))))
+
+      ;; Translate relative paths to group names.
+      (while (not (eobp))
+	(let* ((path (split-string
+		      (buffer-substring (point) (line-end-position)) "/" t))
+	       (art (string-to-number (car (last path)))))
+	  (while (string= "." (car path))
+	    (setq path (cdr path)))
+	  (let ((group (mapconcat 'identity (subseq path 0 -1) ".")))
+	    (push (vector (nnir-group-full-name group server) art 0)
+		  artlist))
+	  (forward-line 1)))
+      (message "Searching %s using find-grep...done" (or group server))
+      artlist)))
+
+;;; Util Code:
+
+(defun nnir-read-parms (query)
+  "Reads additional search parameters according to `nnir-engines'."
+  (let ((parmspec (caddr (assoc nnir-search-engine nnir-engines))))
+    (cons (cons 'query query)
+          (mapcar 'nnir-read-parm parmspec))))
+
+(defun nnir-read-parm (parmspec)
+  "Reads a single search parameter.
+`parmspec' is a cons cell, the car is a symbol, the cdr is a prompt."
+  (let ((sym (car parmspec))
+        (prompt (cdr parmspec)))
+    (if (listp prompt)
+	(let* ((result (apply 'completing-read prompt))
+	       (mapping (or (assoc result nnir-imap-search-arguments)
+			    (assoc nil nnir-imap-search-arguments))))
+	  (cons sym (format (cdr mapping) result)))
+      (cons sym (read-string prompt)))))
+
+(defun nnir-run-query (query)
+  "Invoke appropriate search engine function (see `nnir-engines').
+If some groups were process-marked, run the query for each of the groups
+and concat the results."
+  (let ((q (car (read-from-string query))))
+    (if gnus-group-marked
+	(apply 'vconcat
+	       (mapcar (lambda (x)
+			 (let ((server (nnir-group-server x))
+			       search-func)
+			   (setq search-func (cadr
+					      (assoc
+					       (nnir-read-server-parm 'nnir-search-engine server) nnir-engines)))
+			   (if search-func
+			       (funcall search-func q server x)
+			     nil)))
+		       gnus-group-marked)
+	       )
+      (apply 'vconcat
+	     (mapcar (lambda (x)
+		       (if (and (equal (cadr x) 'ok) (not (equal (cadar x) "-ephemeral")))
+			   (let ((server (format "%s:%s" (caar x) (cadar x)))
+				 search-func)
+			     (setq search-func (cadr
+						(assoc
+						 (nnir-read-server-parm 'nnir-search-engine server) nnir-engines)))
+			     (if search-func
+				 (funcall search-func q server nil)
+			       nil))
+			 nil))
+		     gnus-opened-servers)
+	     ))
+    ))
+
+(defun nnir-read-server-parm (key server)
+  "Returns the parameter value of for the given server, where server is of
+form 'backend:name'."
+  (let ((method (gnus-server-to-method server)))
+    (cond ((and method (assq key (cddr method)))
+	   (nth 1 (assq key (cddr method))))
+	  ((and nnir-mail-backend
+		(gnus-server-equal method nnir-mail-backend))
+	   (symbol-value key))
+	  (t nil))))
+;;     (if method
+;;       (if (assq key (cddr method))
+;; 	  (nth 1 (assq key (cddr method)))
+;; 	(symbol-value key))
+;;       (symbol-value key))
+;;     ))
+
+(defun nnir-group-full-name (shortname server)
+  "For the given group name, return a full Gnus group name.
+The Gnus backend/server information is added."
+  (gnus-group-prefixed-name shortname (gnus-server-to-method server)))
+
+(defun nnir-possibly-change-server (server)
+  (unless (and server (nnir-server-opened server))
+    (nnir-open-server server)))
+
+
+;; Data type article list.
+
+(defun nnir-artlist-length (artlist)
+  "Returns number of articles in artlist."
+  (length artlist))
+
+(defun nnir-artlist-article (artlist n)
+  "Returns from ARTLIST the Nth artitem (counting starting at 1)."
+  (elt artlist (1- n)))
+
+(defun nnir-artitem-group (artitem)
+  "Returns the group from the ARTITEM."
+  (elt artitem 0))
+
+(defun nnir-artlist-artitem-group (artlist n)
+  "Returns from ARTLIST the group of the Nth artitem (counting from 1)."
+  (nnir-artitem-group (nnir-artlist-article artlist n)))
+
+(defun nnir-artitem-number (artitem)
+  "Returns the number from the ARTITEM."
+  (elt artitem 1))
+
+(defun nnir-artlist-artitem-number (artlist n)
+  "Returns from ARTLIST the number of the Nth artitem (counting from 1)."
+  (nnir-artitem-number (nnir-artlist-article artlist n)))
+
+(defun nnir-artitem-rsv (artitem)
+  "Returns the Retrieval Status Value (RSV, score) from the ARTITEM."
+  (elt artitem 2))
+
+(defun nnir-artlist-artitem-rsv (artlist n)
+  "Returns from ARTLIST the Retrieval Status Value of the Nth artitem
+\(counting from 1)."
+  (nnir-artitem-rsv (nnir-artlist-article artlist n)))
+
+;; unused?
+(defun nnir-artlist-groups (artlist)
+  "Returns a list of all groups in the given ARTLIST."
+  (let ((res nil)
+        (with-dups nil))
+    ;; from each artitem, extract group component
+    (setq with-dups (mapcar 'nnir-artitem-group artlist))
+    ;; remove duplicates from above
+    (mapc (function (lambda (x) (add-to-list 'res x)))
+            with-dups)
+    res))
+
+
+;; The end.
+(provide 'nnir)
+
+;;; arch-tag: 9b3fecf8-4397-4bbb-bf3c-6ac3cbbc6664